| 1 |
On Sun, 24 Mar 2013 08:47:41 +0100 |
| 2 |
Ulrich Mueller <ulm@g.o> wrote: |
| 3 |
|
| 4 |
> >>>>> On Sat, 23 Mar 2013, Michał Górny wrote: |
| 5 |
> |
| 6 |
> >> > +multibuild_merge_root() { |
| 7 |
> >> > + [...] |
| 8 |
> >> > + if type -P lockf &>/dev/null; then |
| 9 |
> >> > + # On BSD, we have 'lockf' wrapper. |
| 10 |
> >> > + tar -C "${src}" -f - -c . \ |
| 11 |
> >> > + | lockf "${lockfile}" tar -x -f - -C "${dest}" |
| 12 |
> >> > + else |
| 13 |
> >> > + [...] |
| 14 |
> >> > + cp -a -l -n "${src}"/. "${dest}"/ |
| 15 |
> >> |
| 16 |
> >> You assume that cp is GNU, if there's no lockf. Is it guaranteed that |
| 17 |
> >> this is true on all platforms? |
| 18 |
> |
| 19 |
> > They're all just assumptions. If it breaks somewhere, I'll improve |
| 20 |
> > the function. |
| 21 |
> |
| 22 |
> Maybe explicitly testing for USERLAND would be less fragile? Or use |
| 23 |
> tar everywhere? |
| 24 |
|
| 25 |
I prefer something that can boom into face and let us know that it is |
| 26 |
broken rather than silent, poor catch-all. |
| 27 |
|
| 28 |
-- |
| 29 |
Best regards, |
| 30 |
Michał Górny |
Archives